Академический Документы
Профессиональный Документы
Культура Документы
Simon Kretzschmar
EXAMINER COPY
As the candidates Supervisor I agree/do not agree to the submission of this thesis
Dr J. Pocock
Signature:
Supervisor: Dr J Pocock
I
DECLARATION
(i) The research reported in this thesis, except where otherwise indicated, is my original
work.
(ii) This thesis has not been submitted for any degree or examination at any other
university.
(iii) This thesis does not contain other persons data, pictures, graphs or other information,
unless specifically acknowledged as being sourced from other persons.
(iv) This thesis does not contain other persons writing, unless specifically acknowledged
as being sourced from other researchers. Where other written sources have been
quoted, then:
a) their words have been re-written but the general information attributed to them has
been referenced;
b) where their exact words have been used, their writing has been placed inside
quotation marks, and referenced.
(v) Where I have reproduced a publication of which I am an author, co-author or editor, I
have indicated in detail which part of the publication was actually written by myself
alone and have fully referenced such publications.
(vi) This thesis does not contain text, graphics or tables copied and pasted from the
Internet, unless specifically acknowledged, and the source being detailed in the thesis
and in the References sections.
Signed:
05 / 01 / 2010
II
ACKNOWLEDGEMENTS
This research was made possible by a financial grant from the CSIR in association with Coaltech
2020. I would like to thank my Supervisor Dr J Pocock and Co-Supervisor Prof B K Loveday of the
Mineral Processing Research Group at the University of KwaZulu-Natal for their invaluable
guidance and assistance.
Simon Kretzschmar
06/07/2009
III
ABSTRACT
The mining of coal in arid regions has led to calls for research in to the field of dry beneficiation,
not only for its lower water but also for its lower operating and plant costs. This dissertation
describes coal beneficiation using a dense medium fluidised bed separator developed at the
University of KwaZulu-Natal. The dense medium used being naturally occurring magnetite, a
titanium mining by-product from the Richards Bay region of South Africa.
Initial semi-batch tests were conducted using density tracers followed by batch separation of
discard coal which was in a size fraction of 1.5 to 3.5cm. These semi batch tests allowed for the
characterisation of the bed and the design and construction of a novel separator.
The separation was optimised and tests on the equipment using high ash discard coal under semi
batch operational parameters yielded a separation inefficiency (Ep) of 0.0458 at a split density of
1996 kg/m3 . The 2.5kg batches of coal were fed into the separator and allowed to separate over a
period of 9 minutes. The coal entered at an average ash content of 60.06%. 39.75% of the coal
reported to the floats with a final average ash content of 28.47%. The remaining 60.75% of the coal
reported to the sinks with a final ash content of 80.90%. Continuous operation at a raw coal feed
flow rate of 18 kg/hr yielded an Ep of 0.0462 at a separation density of 1996 kg/m3 . The coal was
fed into the separator at an average ash content of 60.06%. 39.67% of the coal reported to the floats
with a final average ash content of 24.61%. The remaining 60.33% of the coal reported to the sinks
with a final ash content of 76.41%. The experimental data illustrated that dry separation could be
just as efficient as corresponding wet methods (where Ep values of 0.05 are usually obtained).
IV
TABLE OF CONTENTS
CHAPTER PAGE
TITLE PAGE I
DECLARATION II
ACKNOWLEDGEMENTS III
ABSTRACT IV
LIST OF FIGURES IX
1 INTRODUCTION 1
2 PROBLEM ANALYSIS 7
V
3 BASIC PRINCIPLES AND CHARACTERISTICS 14
VI
5.5.3. Bulk Density Variation Within the Bed 39
9 PART 2: CONCLUSIONS 63
10 FUTURE WORK 65
VII
10.1. Scale Up 65
11 REFERENCES 70
12 APPENDICES 72
VIII
LIST OF FIGURES
Figure 14: Schematic of the Construction of the High Pressure Drop Air Distributor 26
Figure 18: Assembly of the Fluidised Bed (Air Distributor Not Shown) 30
Figure 19: Exploded View of the Arrangement of the Fluidised Bed (Air Distributor Not 31
IX
Shown)
Figure 27: Density Tracers Floating on the Surface of the Dense Medium 42
Figure 28: Rear View of the Separator with the Float Removal Chute Removed 45
Figure 29: Front View of the Separator with Feed and Sinks Removal Chutes Removed 46
Figure 30: Front View of the Separator Bed with Feed and Sinks Removal Chutes
47
Removed
Figure 33: Front View of Separator with Feed and Sinks Removal Chutes in Place
50
Figure 34: Back View of Separator with Floats Removal Chute in Place 52
X
Figure 40: Isometric View of the Proposed Design For the Scaled Up Separator 67
Figure 41: Sectioned Isometric View of the Proposed Design for the Scaled Up
67
Separator
Figure 42: Construction Isometric View of the Proposed Design for the Scaled Up
68
Separator
Figure 43: Isometric, Top, Front and Side Schematic Views of the Proposed Design for
69
the Scaled Up Separator
XI
LIST OF TABLES
Table 5: Raw Data Obtained from Batch Test Run with Density Tracers 72
Table 6: Raw Data Showing the Bulk Density Variation within the Fluidised Bed in
74
Plan View (kg/m3)
Table 11: Raw Data Obtained From Continuous Separation of Density Tracers 77
Table 12: Raw Data Obtained From Continuous Separation of Density Tracers 79
Table 13: Raw Data Obtained From Continuous Separation of Density Tracers 81
Table 14: Raw Data Obtained From Continuous Separation of Density Tracers 83
Table 15: Raw Data Obtained From Continuous Separation of Density Tracers 85
Table 16: Raw Data Obtained From Continuous Separation of Density Tracers 87
Table 17: Raw Data Obtained From Continuous Separation of Density Tracers 89
XII
Table 18: Raw Data Obtained From Continuous Separation of Density Tracers 91
Table 19: Raw Data Obtained From Continuous Separation of Density Tracers 93
Table 20: Raw Data Obtained From Continuous Separation of Density Tracers 95
Table 21: Raw Data Obtained From Continuous Separation of Density Tracers 97
Table 22: Raw Data Obtained From Continuous Separation of Density Tracers 99
Table 23: Raw Data Obtained From Continuous Separation of Density Tracers 101
Table 24: Raw Data Obtained From Continuous Separation of Density Tracers 103
Table 25: Raw Data Obtained From Continuous Separation of Density Tracers 105
Table 26: Raw Data Obtained From Continuous Separation of Density Tracers 107
Table 27: Raw Data Obtained From Batch Separation of Coal 109
Table 28: Raw Data Obtained From Batch Separation of Coal 109
Table 29: Raw Data Obtained From Continuous Separation of Coal 110
Table 30: Raw Data Obtained From Continuous Separation of Coal 110
XIII
1 INTRODUCTION
Coal beneficiation offers a considerable number of commercial and environmental benefits. It has the
duel benefit of increasing both the quality and thus value of the coal, but also of allowing the potential
exploitation of coals that would be unrecoverable due to commercial or environmental limitations.
Low ash coals are not only more efficient in terms of combustion, but also result in reduced sulphur
dioxide (which originates from both the organic sulphur and contained sulphides in the coal) and
particulate emissions. These lower emission levels are environmentally desirable. A further benefit of the
removal of ash is the reduction of transportation costs (per gigajoule) which would be due to reduction in
the cost of carrying moist coal.
In the choice of beneficiation techniques, wet coal beneficiation processes are the most popular; this is
due to historically higher separation efficiency and operating capacity: (Cleaner Coal Technology
Programme, 2001). Two main separating principles predominate:
Separation based upon the difference in the relative density of the coal and waste shale; pure coal
has a relative density of approximately 1.3 and associated shale commonly has a relative density
of greater than 2.2. Examples of processes are dense medium separation and jig washing.
Separation based upon the difference in surface properties of the material. Coal is hydrophobic,
while shale is usually hydrophilic. An example of a process utilising this type of separation would
be froth flotation.
1
The wet processing of coal, through dense medium separation or jigs, however requires significant
amounts of water. These can be in the region of 200 litres per tonne of coal processed. (Donnelly J.,
1999)
Figure 1: Mean Annual Rainfall of South Africa (Department of Environmental Affairs and Tourism,
2008)
2
Figure 2: Coal Fields of the Republic of South Africa (Council for Geoscience, 2008)
These deposits, which extend into southern Botswana, are in an area of arid climate conditions, thus the
traditional wet separation processes which have largely been employed in the coal separation industry are
unsuitable. The competition for water is not the only issue as the effluent water from the coal preparation
is generally saline and can be acidic (Donnelly J., 1999).
3
A recent article published in the South African Financial Mail regarding the water crisis in South Africa
only serves to highlight that the situation is reaching a critical level. According to the report, South
Africa, which is 30th on the list of the worlds driest countries, is suffering major effects due to the
pollution of her main rivers and water supplies. One of the main sources of contamination is effluent from
mining operations. In fact one of Gautengs major water sources, the Vaal River, is showing increased
pollution levels from acid mine water. Further north in Mpumalanga, the Olifants River system which
runs through the coal preparation units and supports many farms in the area, is also being affected by
toxic mine water (Financial Mail, 2008). These factors are of considerable worry to a country that
depends on coal for not only 90% of its electricity and 30% of its petrol and diesel but also 90% of its
iron and steel production. Added to this are substantial revenues from foreign exchange brought in by
coal exports. (Creamer M., 2009)
Dr Andrew Turton, in his paper; Three Strategic Water Quality Challenges that Decision-Makers Need
to Know About and How the CSIR Should Respond, states in no uncertain terms (Turton A., 2008):
South Africa simply has no more surplus water and all future economic development (and thus social
wellbeing) will be constrained by this one fundamental fact...
4
1.3. DRY BENEFICIATION: AN ALTERNATIVE
Dry beneficiation has several notable advantages. The principle of these is that the elimination of the need
for water obviously eliminates the need for expensive dewatering processes such as pumping, screening,
filtering and centrifuging. The saline and acidic water from the wet processing would need further
treatment, an additional cost as would the removal of entrained fines. Finally freight costs per gigajoule
will be considerably lower due to reduction in the cost of carrying moist coal. Thus the dry coal
preparation plants would experience the multiple advantages of being smaller, cheaper, and having lower
operational costs.
Traditionally however the disadvantages of dry preparation have outweighed these advantages. Coal
cleaned in dry processes generally have ash contents that are higher than that of coal cleaned in more
efficient wet methods, and while dry processes are susceptible to feed moisture content. This issue is not
experienced with wet preparation. The problem of dust formation with dry processing is always present as
is the difficulty of dry screening. Dry processing equipment on average tends to have lower capacities
than corresponding wet methods. These disadvantages, if recognised, may however be overcome
(Donnelly J., 1999).
5
1.4. AIMS AND OBJECTIVES
At this point the primary objectives of the project became clear, they were:
To identify a method of dry separation, that allows for accurate, high capacity separation without
excessive dust formation.
Develop the equipment, such that separation can be undertaken under continuous operating
conditions.
Investigate the scale up potential of the equipment to pilot plant capacity and beyond.
6
1.5. PROJECT LOGIC FLOW
A logic flow was determined for the project to facilitate the most effective method of proceeding with
the design and development of the equipment.
7
2. LITERATURE REVIEW
Wet coal beneficiation techniques have come to represent the standard in coal beneficiation techniques;
they provide a combination of accuracy of split and high capacity operation that the traditional methods of
dry beneficiation could not compete with. Furthermore the moisture content of the coal is not a factor to
be considered before beneficiation can take place. Techniques vary depending upon the size of the run of
mine coal sent to the beneficiation process, however it is noted that the general separation inefficiency Ep
for wet processing is between 0.007 (ESR International 2010) to 0.015 (Portaclone 2010).
Processing of wet coal generally follows a flowsheet typical to the one in figure 4. The coal from the mine
is crushed to a top size that is acceptable through breakers, mills and crushers. From there the coal is
screened into size fractions. Based upon these size fractions the coal will beneficiated according to the
different methods described below.
Figure 4: Coal Wet Processing Flowsheet (Chiang S.H. and Cobb J. T., 1993)
Coarse Coal
Two methods generally predominate, dense medium separation and jig washing. Two actions comprise
the jigging operation. The first is the effect of hindered settling. This results in the lighter particle settling
slower than the heavier one. The second separation process is achieved by an upward flow of water which
8
segregates the particles by density. By means of slurry pulses, these two actions are combined in the jig.
Gravity separation utilises the settling rate of different particles in water to make a separation. Particle
size, shape and density all affect the efficiency of the separation.
Dense media separation takes place in fluid media with a density between that of the light and heavy
fractions that are to be separated. The separation is dependent upon density only (Metso, 2009). Dense
medium separators have low separation inefficiencies generally with an Ep value in the region of 0.002
(Bateman Engineering, 2009).
The choice between these methods comes down to the unique circumstances innate to each plant. Dense
medium washing provides a more accurate split, however jig washing is often perceived to be a simpler,
lower cost option when accuracy of split is not paramount.
Beneficiation of coal in this size fraction is becoming of greater importance and is usually
accomplished either through spiral concentrators or teetered bed separators. These separators
again achieve the split between the heavy and light fractions based on the density of the
constituents.
Froth flotation is still the most widely used method of beneficiation for coal of
this size as density separation becomes difficult for smaller particles. This physiochemical
process involves the selectivity of the attachment of air bubbles to the organic coal particles and
not the not the surrounding non-organic minerals. The coal is made hydrophobic by the addition
of a surfactant and an oil is used as an agglomeration agent to allow easy removal of the froth.
(Davydov M.V. 2008)
The coal that has undergone the flotation separation commonly has unacceptable amounts of water and
thermal drying is usually used to reduce this moisture content.
9
After the coal has been processed the treatment of the water used for the beneficiation becomes a focus,
and the main disadvantage of wet coal beneficiation comes to the fore. After the coal has been processed
the larger fraction products and rejects from the jigs and dense medium separators are rinsed. The slurries
from the fine coal processing still need to undergo filtering and other dewatering processes such as
settling ponds (Chiang S.H. and Cobb J. T., 1993). The treatment of the tailings and water clarification
have and still remain the most costly areas of coal beneficiation, and at the same time the most difficult to
control. Whilst significant attempts are made to reduce the amount of water that is actually required 200 L
of water per ton of coal are still lost through product coal moisture, disposal of the tailings and
evaporation. (Donnelly J., 1999). Where the buildup of slimes in the water reaches a maximum the waste
water is disposed of in tailings ponds where the evaporation of the water leaves the recovery of the ultra
fines possible. (Clark K. 1997).
For these processes there is the need for a high level of control in the area of online process monitoring. A
trend over recent years is now to develop simpler processes that utilise larger single separation units. This
trend together with a desire for plants that are modular and allow for quick relocation are not traits
generally inherent to the traditional wet coal processing operations. (Cleaner Coal Technology
Programme, 2001).
10
2.2. DRY SEPARATION TECHNIQUES
The first dry cleaning methods of coal involved the removal of waste by handpicking off slow moving
conveyers. Over time these methods were refined into todays technologies mentioned below. Despite
growth during the first half of the 20th century, dry beneficiation processes were abandoned. The reason
for this lay in the fact that available technology restricted the feed size, capacity and coal moisture content
that the separators were able to a cope with. These restrictions combined with inaccurate separation meant
that the popularity of dry coal cleaning fell off dramatically compared with wet separation techniques.
However in areas where water is particularly scarce, such as in regions of China, dry separation methods,
mentioned below, may still be found. (Donnelly J., 1999)
Fluidised bed dry cleaners, became generally viewed as the most productive means of dry separation
processes. They include pneumatic oscillating tables, air jigs and dense medium fluidised bed separators.
Most common of these were the oscillating air tables.
In the oscillating air tables the oscillation of the coal bed together with fluidising pulses of air from
beneath the bed allowed vertical stratification of the coal, fractions were taken off along the length of the
table through various skimmers. While the tables operated at relatively high separation inefficiencies,
clean coal with an ash content of 10% and reject coal with an ash content of 70% could be achieved.
Explosive dust clouds, caused by the dry coal, were usually controlled through water mist sprayers. By
the 1960s however oscillating air tables had fallen out of favour, mainly due to their inability to process
the high ash and high moisture coals that were being increasingly encountered. Besides deshaling
operations such as selective crushing dry beneficiation was virtually non-existent. (Donnelly J., 1999)
11
Figure 5: Diagrammatic View of Oscillatory Air Table (Donnelly J., 1999)
In a jig, an eccentric drive located at the feedbox head serves to impart an up or down motion to the
jigging box. This motion is controlled by a fixed fulcrum located at the discharge end of the box. The
fulcrum decreases the motion of the box from a maximum to minimum in the direction of feed to
discharge. In addition to this a pulsating air current is applied through an air distributor screen located on
the base of the jigging box. The pulse is applied during the downward stroke of the beds jig in order to
fluidise the contents of the bed. The resulting stratification of the bed occurs with the high density
material settling to its base and the lower density coal rising to the surface. The high density rejects are
discharged by means of an adjustable toothed roller while the clean coal is removed via a chute. Due to
significant dust formation such equipment is generally fitted with an extraction system.
While still on a decline air jigs are still found in some areas. An example of an air jig dry separation
process is the FGX series compound dry cleaning machine. This is a series of machines that have found
growing popularity over recent years, with a maximum capacity of 480tph, an efficiency of 90% and the
ability to process coal with a surface moisture of up to 9% these machines are a step in the positive
12
direction especially considering the investment costs are up to a tenth that of that of similar capacity wet
plants (Tangshan Shenzhou Machinery Co., Ltd, 2009). Yet still, the potential use of such machines is
limited by their operating capacities and separating efficiencies which relative to wet processes remain
low. In spite of extraction systems and dust enclosures the problem of dust generation still remains. Most
dry separators require narrow bands of coal sizes and this in itself requires significant work to achieve.
Despite continued investigations the best achievable separation inefficiencies by pneumatic equipment are
in the region of 0.3 (Industrial Technologies Program 2006). Higher efficiency in the fluidised bed
separation has been achieved using dense media separation.
Figure 6: FGX Cleaner Schematic (Tangshan Shenzhou Machinery Co., Ltd, 2009)
Before the investigation into the dense medium beds is undertaken it is worthwhile to look at several
other points to note in the modern world of dry separation.
The advantage of the reduced size of the dry coal processing units compared with their wet counterparts
has been noted especially with regards to the potential to have mobile processing units that can operate
near the extraction point of the mining operation (Industrial Technologies Program 2006).
The magnetic properties of the raw coal also lead themselves to exploitation in dry separation methods.
The organic coal is diamagnetic and the pyrite in the coal is paramagnetic, with weakly magnetic
13
constituents in the ash also predominant. The most well known examples are High Gradient Magnetic
Separation (HGMS) (Chiang S.H. and Cobb J. T. 1993) and triboelectric separation (see figure 7) (Qing-
ru C. and Hai-feng W. 2006).
Whilst these processes are able to achieve high ash reduction down to 8% with yields of nearly 80%, the
coal does need to be pulverized to a significant degree prior to processing (0.043mm). These applications
are mainly used in the factory boilers and iron smelting blast furnaces. (Qing-ru C. and Hai-feng W.
2006)
The medium (similar to that of a magnetite medium in a wet plant) is an air-solids mixture. Then mixture
provides a medium that is stable and of uniform density. Materials of higher density (such as ash) sink,
14
while lower density coal floats. This provides a quick efficient separation. The schematic of the separator
is shown in Figure 8.
Compared to air jigs and tables the pressure and volume of air required is lower and the smaller amount
of dust produced is more easily dealt with. Separation of the sinks and floats is achieved by a push plate
chain conveyor that scrapes the floats off one end of the machine whilst removing the sinks from the
other. The results indicate that 6mm coal can effectively be separated at an Ep of 0.05 (where the Ep
value represents the degree of inaccuracy in the split between the sinks and the floats at the separation
density). A gas-solids fluidised bed with a uniform and stable density can be formed utilising a magnetite
powder or a mixture of powder and fine coal. Using tight control on the fluidization and bed composition
a separation density range of 13002200kg/m3 is achievable. The advantage of this technology is that the
costs associated with construction and operation are about half those for similar scaled wet processes.
This is in addition to lower environmental impact (Zhenfu L. and Qingru C., 2001).
The advantage of dense medium separation is the flexibility of the design of the process. The Reflux
Classifier with its parallel inclined channels that are situated above a fluidized bed uses a combination of
vibration and a dense medium of sand to achieve the coal beneficiation. Separation inefficiencies Ep of
0.07 have been achieved whilst realizing a reduction of the ash in the products to 15% with a 80% yield.
(MacPerson S.A. et al 2009)
15
3 BASIC PRINCIPLES AND CHARACTERISTICS
Dry fluidisation is the levitation of a bed of solid particles by a gas. The bed, in this levitated state,
exhibits fluid like behaviour (as shown in Figure 9). Thus it tends to establish a level and flow in response
to pressure gradients that may be present in the bed (Pell M., 1990).
Figure 9: Pseudo Fluid Characteristics of Gas Fluidised Solids (Zhenfu L. and Qingru C., 2001)
16
3.2. PRINCIPLES OF FLUIDISATION
The principles of fluidisation and their associated properties now come into prominence as the careful
fluidisation of the dense medium needs to be considered. The fluidisation regime needs to produce a
dense medium that is uniform with a consistent predictable density.
Unlike liquid fluidised systems which present steady predicable behavior, the behavior of gas fluidised
systems are far more complicated. Much of this complication arises from the interaction of the frictional,
electrostatic and surface forces between very fine particles, which have a far greater effect than the
hydrodynamic forces otherwise experienced. Coulson and Richardson (2002) describes the stages that a
system experiences with an increase in gas velocity:
Fixed Bed
In this state, until the velocity has been increased to such a point where the pressure drop across
the bed is equal to the weight per unit area of the particles in the bed, the particles remain in
contact with each other resulting in a stable bed structure. This is the point of incipient
fluidisation, and is velocity that of umf, the minimum fluidising velocity.
Particulate Fluidisation
The bed now begins to expand as the velocity increases. And although the agitation experienced
by the particles increases the bed maintains uniformity. This type of fluidisation is typical to
liquid fluidisation. Gas solid fluidisation usually only experiences this type of fluidisation at very
low velocities and in some cases not at all before bubbling begins.
17
Aggregative Fluidisation
This fluidisation, also known as bubbling fluidisation, is characterised by the formation of two
separate phases, a dense phase made up mainly of solids and a discontinuous lean phase formed
by the channeling of the gaseous fluid phase through the particles.
Turbulent Fluidisation
This is a chaotic region in which the bubbles coalesce and their identity is lost.
Fast Fluidisation
This condition generally lies outside the realm of true fluidisation and represents the stage where
there is transport of the particles vertically upwards.
Figure 10: Fluidisation Regimes (Adapted from Perry R.H. and Green D., 1998, Perrys
Chemical Engineers Handbook)
18
The processes of feedstock variation and attrition usually result in a mixture of particle sizes in a fluidised
bed. A mixed size bed fluidises more smoothly than a closely sized one. The smaller particles fit
between than larger ones and act as ball bearings or a lubricant to make flow easier. A range of particle
sizes spanning an order of magnitude is reasonable. (Pell M., 1990)
The properties of the particles determine the ease with which fine particles can be fluidised. Whilst the
nature of the fluidisation for a particular group of particles is exactly predictable it is possible to view
trends in the fluidisation. Geldart (Geldart D., 1973) classified particles into four groups. Coulson and
Richardson (2002) provide Table 1 below, grouping the particles and Pell (1990) provides their location
in a particle density size chart shown in Figure 11.
19
Figure 11: Geldarts Characterisation of Fluidisation Regime
The pressure drop across the bed is plotted against increasing fluidising gas velocity; a characteristic
curve (Figure 12) is produced. Coulson and Richardson (2002) description is as follows.
As the gas velocity increases the bed begins to expand linearly from rest to point (A), at this point the
particles within the bed become rearranged and the slope of the curve begins to decrease. As the
aerodynamic drag forces begin to counter the gravitational forces the bed begins to expand as the particles
move away from each other. When this aerodynamic drag equals the gravitational force the particles
become suspended within the bed the pressure drop passes through its maximum at (B) and moves
towards steady state conditions (C to D). No further increase in gas velocity at this point will produce a
change in bed pressure drop. It should be noted that the straight line region (from rest to point A) is the
packed bed region. This is where the particles do not move relative to one another and their separation is
20
constant. In this region the pressure drop versus velocity relationship can be described by the Ergun
equation.
It may be noted that should the gas velocity be reduced at this point the characteristic double pathway (E
to F) can be formed (as seen in Figure 10). This is due to the lower pressure drop typical of an expanded
reformed bed resultant from settling particles. Should there be a vibration present there is a good chance
that this expanded bed would not occur, as the particles are forcefully settled into their original more
compact state. Plotting of the pressure drop versus the fluidising velocity is usually conducted on the
logarithmic curve as shown below, however it may be conducted on a semi-logarithmic or even linear
depending on the nature of the system.
The minimum fluidising velocity umf may now be determined by experimentally measuring these points,
plotting them and then using straight line plots through the EF and CD sections to find their intersection.
Figure 12: Pressure Drop over Fixed and Fluidised Beds (Coulson J.M. and Richardson J.F., 2002)
Where the y axis represents the the pressure drop over the bed and the x axis the superficial fluid
velocity throught the bed.
21
3.3 DENSE MEDIA
In traditional wet dense media separation a suspension of dense powder in water is used to form a Dense
Liquid (a suspension of specified density according to the volume fraction of fine solids) in which the
sink-float separation process takes place. However the formation of a stable, uniform gas-solid dense
medium is far more difficult and requires the careful combination of correct particle size, fluidising gas
velocity and uniform air distribution. The large scale instabilities and the general heterogeneous nature of
a gas-solid dense medium is in contrast to the homogeneous fluidised bed typical of a liquid-solid dense
medium. These homogenous beds are a result of the higher viscosity of the fluid and similar intrinsic
densities of the solids and fluid (Kunii D. and Levenspiel O., 1991). The bulk density of the gas-solid
dense medium and thus the split density of the bed is a function of the density of the particles and the
voidage between them (determined by the degree of fluidisation).
..When a body is completely or partially immersed in a fluid, the fluid exerts an upward force on the
body equal to the weight of the fluid displaced by the body....
In essence a body of a lower density will float on the surface of a more dense liquid, whilst bodies of
higher density will sink.
Dense media separators draw on this principle and the fact that the dense medium behaves like a liquid
when fluidised. The coal containing sulphur in the form of pyrites and ash is heavier than the desired
high carbon coal and thus a separation density is established. With the knowledge of the separation
density a fluidised bed can be created so as to separate the unwanted coal (containing more ash and
sulphur) from the desired coal. (Zhenfu L. and Qingru C., 2001)
22
The determination of the separation inefficiency in a dense medium separator is obtained from the
partition curve and is reflected by the partition coefficient. (Zhenfu L. and Qingru C., 2001)
75 25
Ep (1)
2000
Where the symbols 75 and 25 represent the densities (in kg/m3) through which 75% and 25% respectively
of the feed report to the underflow. This Ep value thus represents the inaccuracy of the split between the
low density desired high carbon content coal and the undesired higher density high shale and ash material.
Figure 13, below, shows the difference between an ideal split and a typical density separation curve. The
Ep described in Equation 1 would be 0 for the ideal split as all the material above the split density reports
to the underflow and all the material below reports to the overflow. The non-ideal curve shows a truer
representation of what actually occurs with the effects of entrainment, particle shape, particle interaction
and middlings coming into play.
Having considered the principles of fluidisation, careful attention now needed to be paid to the
construction of the dense medium bed in which the separation would take place and within which the
separator itself would be placed.
The design of the dense medium fluidised beds revolves mainly around the medium to be fluidised, as the
difficulty of uniform gas-solid fluidisation is notorious. The dense medium that was chosen for use was
magnetite (Fe3O4). This black ferromagnetic mineral has been used in previous coal dense medium
separation applications (Zhenfu L. and Qingru C., 2001) and is common to wet dense medium
separations. It has a density of approximately 5150kg/m3 (Excalibur Mineral Company, 2008) and a bulk
density of approximately 2450kg/m3 (calculated from experimental analysis of weights of known
volumes).
The magnetite obtained for the dense medium was a by product of the titanium beach mining operations
on the northern KwaZulu-Natal coast. This material has experienced attrition by wave action before it was
deposited on the beach. This attrition produces a material that has a size spectrum that is narrow enough
for uniform fluidisation, yet still includes enough fines to promote fluidisation (see Figure 14).
24
Figure 14: Size Distribution of Magnetite
The magnetite offers a bulk density that is ideal for coal beneficiation and has the added advantages of
allowing magnetic recovery from the coal product.
The even fluidisation of the dense medium is of critical importance to the accuracy of the split within the
bed. The most important determining factor that needs to be controlled is the air distribution into the bed
of particles. The fluidisation regime of air fluidised magnetite, predicted according to Geldarts
classification mentioned in the previous section, is that of a bubbly bed. These Type B particles do not
form a cohesive structure that allows for uniform expansion; rather once the minimum fluidising velocity
has been exceeded the formation of bubbles occurs. Poor air distribution will result in the formation of
bubbles in some areas of the bed, whilst others are incompletely fluidised.
25
4.2. DESIGN OF FLUIDISED BED EQUIPMENT
The air distributor also serves the dual purpose of supporting the weight of the bed. Due to the bubbling
action of the fluidised bed, there are constant small changes in the local pressure drop at the distributor.
Gas inevitably tries to enter the bed in the zone of lowest pressure drop. The distributor pressure drop
therefore has to be large enough to overcome the small local pressure disturbances of the bed. If the
pressure drop is too small, gas will end up flowing through only some portions of the bed and establishing
flow paths with high voidage and low pressure drop. At the same time, other sections would have
negligible flow and remain closed or non-bubbling.
The above criteria are key the design of the distributor. In practice the required pressure drop is set by
operating experience. For up flow the design pressure drop should be at least 30% of the bed pressure
26
drop, dp, at the minimum expected gas flow and the maximum expected bed weight (Pell M., 1990). A
comfortable margin of safety is to design the grid dp for 100% of the bed dp (Coulson J.M. and
Richardson J.F., 2002). This is often no strain for systems in which the gas is coming from a relatively
high pressure source. The problems associated with this are that, while the gas will be well distributed, the
distributor and plenum need to be designed to take these pressures. For drilled plate distributors the
velocity of the gas through the holes might become high enough to cause an attrition problem (Pell M.,
1990).
For the design in question it was decided to use a high pressure drop distributor considering the bubbly
nature of the bed in order to ensure there was even air distribution into the bed. The distributor was
constructed using filter cloth sandwiched between two stainless steel metal grids. This pressure drop was
high enough to allow for uniform distribution of air whilst at the same time providing a stable structure on
which the bed could rest. The metal grids prevented the ballooning of the pressure cloth either from the
weight of the bed at rest or the force of the air being forced through during operation (see Figure 15 and
Figure 16).
Pressure Cloth
500mm
Figure 16: Schematic of the Construction of the High Pressure Drop Air Distributor
27
4.2.2. The Plenum
The design of plenum is almost of as critical importance as that of the bed itself. Correctly done it can
provide additional air distribution benefits. Pell (1990) suggests the alternative plenum shapes shown
below in Figure 17, (Pell M., 1990). The final chosen design was based on trial and error experiments.
The plenum consisted of a segregated honey comb structure Figure 18 beneath which there was a
tetrahedral constriction towards the air supply flange from the blower Figure 19. The constriction allowed
for good initial air distribution whilst the honey comb structure created not only a secondary air
distribution zone, but also a support structure upon which the distributor could be fastened (Figure 20).
Both the primary and secondary plenums were constructed from PVC plastic.
28
Figure 17: Possible Plenum Arrangements (Pell M., 1990)
29
Figure 18: Schematic of Secondary Air Distributor
30
Figure 20: Assembly of the Fluidised Bed (Air Distributor Not Shown)
31
4.2.3. The Bed Walls
The fluidised beds walls were made of interlocking sections that allowed for bed height variation from
100mm in height to 300mm in height, to allow for sufficient separation to occur. These walls were
constructed from clear Perspex to allow for a degree of visual inspection of the fluidization (see Figure 21
below).
100mm
450mm
450mm
500mm
150mm
1000m
m
Figure 21: Exploded View of the Arrangement of the Fluidised Bed (Air Distributor Not Shown)
32
5 BED FLUIDISATION ANALYSIS
Initial tests on a laboratory bed (see Figure 22) were conducted on the magnetite to analyse its fluidisation
characteristics. The fluidisation characteristics allow prediction of the nature of fluidisation that the
magnetite would experience in the scaled up pilot plant bed and in addition allowed the testing of the high
pressure drop filter cloth.
The first set of fluidisation tests were conducted in the laboratory bed in Figure 1. The bed was filled with
magnetite to the same depth that would be used in the pilot plant bed, namely 20cm. The bed was
equipped with flow rotameters to regulate the air fed to the bed and a manometer to determine the bed
pressure drop.
The effect of the distributor on the uniform fluidisation of the bed was tested. Initially a perforated plate
was used; however this did not prove to be effective at all in providing uniform air distribution. The bed
experienced channeling in its center, whilst at the walls dead zones were found. It was thus necessary to
increase the pressure drop across the distributor to improve distribution. A substantial improvement in the
air distribution was found by replacing the perforated plate with the high pressure drop distributor
described previously. The results of the fluidisation would show under what conditions the magnetite
would fluidise, what the nature of that fluidisation would be, its stability and the bulk density at which
separation would occur.
33
Manometer
Flow
Rotameters
Dense Medium
(silica/magnetite
in this view)
Distributor
Plenum
34
5.3. FLUIDISATION CHARACTERISTIC CURVE
The characteristic fluidisation curve of the magnetite can be seen in Figure 23. This curve is typical of the
fluidisation of fine particles. The gas velocity at the point of insipient fluidisation was 0.165m/s. The
presence of fine particles within the bed aided fluidisation. The minimum bubbling point, as is typical of
Geldart classification B particles, was experienced at 0.17m/s. The excellent distribution of air from the
high pressure filter ensured that the fluidisation was uniform throughout the bed.
35
Gas Velocity, u (m/s) 0.00 0.04 0.06 0.08 0.10 0.11 0.13 0.17 0.18 0.20 0.23 0.35
Bed Pressure Drop (kPa) 0.00 13.89 21.65 30.49 39.52 46.91 59.36 79.39 78.85 79.39 86.25 92.20
Bed Height (cm) 20.00 20.00 20.00 20.00 20.20 20.30 20.70 21.10 23.00 23.10 23.10 24.00
Bed Density (kg/m3) 2200.73 2200.73 2200.73 2200.73 2173.24 2159.49 2104.51 2049.52 1788.34 1774.60 1774.60 1650.88
36
5.4. MAGNETITE BED BULK DENSITY
The bulk density of the static bed was initially determined by observation of its volume relative to its
known mass, and the resulting bulk density during fluidisation was then determined by the beds
expansion. The large drop in bed density at the superficial gas velocity of 0.17m/s was due to bubble
formation. Analysis was then conducted using density tracers to determine whether the bed was fluidised
enough to allow separation. Separation was limited by excessive bubbling within the bed. This excessive
bubbling resulted in the bed acting more like a mixer than a separator. Effective separation was found to
exist between a bulk density of 1900 and 2050kg/m3.
37
5.5. INITIAL SEPARATION TESTS
Initial separation tests were conducted on the laboratory scale pilot plant. Density tracers were
constructed to demonstrate the density split. These PVC cylinders (20mm in diameter and 20mm in
length) were filled with lead and ferrosilicon in order to achieve the correct density for each fraction that
they represented. The density spectrum covered by the tracers is shown in Appendix 4. Three tracers were
constructed for each density fraction to reduce the possibility of entrainment errors and increase accuracy.
The density spectrum of 1350 to 2850kg/m3 serves to represent the mixture of coal and shale that would
be received from the mine.
38
5.5.2. Batch Split Tests
The initial analysis of the pilot plant bed involved a sink-float test. The 45 density tracers (15 fractions
with 3 tracers per fraction) were loaded into the bed that was operating at a superficial air velocity of
0.17m/s. The tracers were allowed 20 seconds to segregate then the air supply to the bed was stopped.
This ensured that, as the tracers were being removed by hand, there would be no movement of the sinks
and floats from their density fractions. The separation inefficiency of the tests can be seen below by the
partition curve in Figure 26, with a split density of 2025kg/m3. This split density was determined at the
50. This is the point where the tracers would theoretically have a 50% probability of reporting to the sinks
or floats fraction. This point would correspond to the bulk density of the bed.
Split density at 50
39
5.5.3. Bulk Density Variation Within The Bed
Analysis of the density variation within the bed was then conducted. The bed was subdivided into 25 test
zones. In each of these test zones a sink float test as described above was conducted to determine the local
density. The sink float tests were done in preference to using hydrometers due to the effect of the bubbles
within the bed on the hydrometer. The results of the density distribution can be seen in Figure 27 below.
It is important to note that whilst there were observable wall effects, these were minimized through
correct bed design. This resulted in a variation of approximately 4% in the bulk density of the bed.
40
6 PART 1: INITIAL CONCLUSIONS
The fluidisation of the magnetite successfully produced a uniform medium. The presence of minor
bubbling in the bed (see Figure 28) did not affect its separating capabilities. The bubbling in fact aided in
separation by providing a jigging action to the bed. This liberated the lighter tracers that were entrained
beneath heavier ones sinking to the bottom of the bed.
The uniform air distribution below the bed provided by the segregated plenum and the high pressure drop
distributor cloth ensured that there were few noticable dead zones in the bed. Due to the density and size
fractions of the magnetite used there was little in the way of dust formation from the fluidisation. Any
loss of magnetite from the bed came from the eruptions of the bubbles, however this was minimal.
41
DENSITY SEPARATION WITHIN THE BED
The density split separation provided by the bed was exceptionally accurate with a separation ineffiency
(Ep) of 0.015 at a bulk density of 2023kg/m3. In fact of the four runs conducted with the density tracers
only one tracer from the floats fraction reporting to the sinks and one from the sinks reported to the floats,
even so these tracers came from a density fractions immediately above and below the split density (See
Figure 26 in Section 5 and Table 5 in Section 12.1 Appendix 1).
Figure 29: Density Tracers Floating on the Surface of the Dense Medium
Density tracers from the floats fraction in the fluidised bed are shown above in Figure 29. There was an
increased bubbling effect in the centre of the bed caused by the presence of the sinks at the distributor.
These sinks created a zone of lower pressure increasing the air flow at this point. The resultant bubbling
meant that during the separation the bulk density of the bed was lower toward the centre of the bed.
42
This variation of bed density whilst having no effect on the actual separation would affect the efficiency
of the control of the process. Thus the separator would need to ensure that there was no interaction either
between the sinks or the separator itself with the distributor. It would, in addition, need to provide a
mixing action within the bed that would break up the formation of large bubbles and dead zones.
43
7 SEPARATOR DESIGN AND CONSTRUCTION
1. Provide a mechanism for segregation and continuous removal of floats and sinks within
the bed.
2. Not disrupt or adversely affect the fluidisation of the dense medium within the bed. There
would need to be adequate clearance between the separator and the air distributor. Initial
test work indicated that the presence of sinks in close proximity to the distributor
produced excessive bubbling.
3. Provide an adequate degree of mixing that will prevent the formation of dead zones
within the fluidised dense medium.
44
7.2. FINAL DESIGN CHOICES
The final separator that was designed and constructed was a rotating basket that would be partially
immersed in the bed.
Outer Basket
Central Weir
Angled Lifters
Steel Support
Structures
Figure 30: Rear View of the Separator with the Float Removal Chute Removed
The outer basket, 450mm long with the same diameter, was constructed of stainless steel mesh. This mesh
was of 1mm diameter with 9mm gaps between the bars. The cylindrical structure of the outer basket was
reinforced by 8mm steel rod (See Figure 30). The spacing was chosen after numerous experimental tests.
It provided a rigid support structure that would maintain the integrity of the separator whilst allowing free
flow of the fluidised magnetite. The sizing of the gaps needed to be minimal in order to prevent the
density tracers and, in later tests, pieces of coal from slipping between the gaps. These particles would
accumulate on the surface of the distributor, destroying the beds stability.
45
7.2.2. Separator Sub Division
The basket would be segregated into two sections by means of a solid plate weir, one for the removal of
the sinks and the other for the removal of the floats. The solid weir would provide additional stability to
the central section of the separator. The weir would prevent the sinks from slipping into the floats removal
section while the floats could easily pass over the top.
The unique internals of the separator provided a guidance mechanism to promote the segregation of the
sinks and floats (see Figure 31).
Mechanical
Scrapers
Support Shaft
Drive Cog
Drive Chain
Figure 31: Front View of the Separator with Feed and Sinks Removal Chutes Removed
46
Figure 32: Front View of the Separator Bed with Feed and Sinks Removal Chutes Removed
47
7.2.3.1. Floats Segregation Corridor
The segregation corridor in the sinks removal section was suspended from the extended drive shaft. The
feed material was fed directly into this corridor. The walls extended below the surface of the fluidised
dense media, thus the floats were able to move down the corridor, over the weir and into the floats
removal section without being caught in the lifters in the sinks section. The sinks simply sank to be
collected by the rotating lifters (see Figure 33).
Drive Cog
Weir
Support Shaft
Support Arms
Floats
Segregation
Corridor
48
7.2.3.2. Mechanical Scraper
Scraper Shaft
Scraper Action
Scraper Head
The passage of the floats was aided by a mechanical scraper. The scraper was manually operated and was
controlled from the floats removal side of the separator. It extended from wall to wall in the Floats
Segregation Corridor and moved back and forward towards and away from the weir (see Figure 34, which
was also mesh based so as to cause as little disruption to the fluidisation).
49
7.3. SEPARATION MECHANISM
The coal was fed into the sinks removal section of the separator.
The feed was constrained in sinks removal section so that floats were not removed with the
sinks by the lifters at the separator walls.
Feed Chute
Front End
Plate
Sinks Removal
Chute
Figure 35: Front View of Separator with Feed and Sinks Removal Chutes in Place
50
The sinks sank to the floor of the separator where they were scooped out by the lifters
attached to the wall of the rotating basket.
The sinks roll down the angled lifters and out of collection chutes located above the beds
horizontal axis (see Figure 35).
The floats are transported past the dividing weir to the floats collection section of the
separator by means of a mechanical scraper.
In the floats collection section the floats are dragged to the walls by means of a current
induced in the bed by the motion of the separator. Once at the walls they are scooped up by
angled lifters attached to the separator wall and removed via a product chute in a similar
manner to that of the sinks (see Figure 36).
51
Floats
Removal
Chute
Free Area
Below
Separator
High Pressure
Distributor
Cloth
Figure 36: Back View of Separator with Floats Removal Chute in Place
52
7.4. AIR SUPPLY AND CONTROL
The air supply to the separator was provided by a high volume blower. A geared motor drove the
separator. The blower settings were manipulated in the control hut (see Figure 37).
Control Hut
Separator
Drive Motor
Air Blower
53
8 PART 2 : OPTIMISATION AND OPERATION OF THE NOVEL
SEPARATOR
With the separator constructed test work was conducted to determine its optimum operating point. From
that stage the batch and continuous operating characteristics would be analysed using density tracers and
coal.
The density tracers were again used for initial characterisation of the separator providing a feed that was
both uniform in size and shape, and was easily classifiable. This allowed the number of runs undertaken
to be maximised. In the optimisation tests a sample of tracers was fed into the separator. The rotational
speed of the separator and the air flow rate to the bed was varied. The resultant separation inefficiency
(Ep) was noted and averaged for 3 tests at each of the conditions. The optimum conditions can be
observed from the bar plot Figure 38. They show a peak efficiency achieved at a superficial gas velocity
of 0.13m/s and separator speed of 4rev/min.
54
Figure 38: Composite Optimisation Graph
55
8.2. BATCH ANALYSIS OF TRACER SEPARATION
These optimum conditions allowed initial separations to be carried out on the equipment. Again density
tracers were used for ease of post separation analysis. The results of this separation can be seen in Figure
39. These results are the combined analysis of 15 batch tests. In each batch test 2.5kg of density tracers,
of an even density distribution, were fed into the separator over a period of 9 minutes and allowed to
separate. An average separation inefficiency (Ep) of 0.057 was achieved at a split density of 1871kg/m3.
56
8.3. BATCH ANALYSIS OF DISCARD COAL SEPARATION
The next investigation into the efficiency of the separator was conducted using high ash Waterberg
discard coal, crushed and screened to a size fraction of 1.53.5cm. Again the coal was fed in to the
separator under the same conditions as the tracers had been in previous tests. This coal had very low free
moisture content as it was dry to the touch.
After collection of the floats and sinks the samples were separated into their respective density fractions
through dense liquid sorting. The separate fractions were then crushed and combusted in order to
determine their ash content. The results of the separation can be seen in Table 3. The feed material,
having an ash content of 60% w/w was separated into a floats fraction having an ash content of 28% and a
sinks fraction having an ash content of 80% w/w. The Partition curve obtained can be seen in Figure 40.
The same operating conditions were used as with the density tracers. A split was achieved at a bulk
density of 1996kg/m3 with a separation inefficiency of 0.046 obtained.
57
Figure 40: Partition Curve for the Batch Beneficiation of Coal
58
Density Floats Overflow/Feed Sinks Overflow/Feed Calculated Partition
Fraction Overflow Ash (%) (%) Overflow Ash (%) (%) Feedstock Feed Ash (%) Coefficient
3
(kg/m ) (%) (%) (%) (%)
<1400 16.82 8.59 6.56 0.002 16.34 0.001 6.57 8.59 0.02
1400-1500 31.76 10.35 12.39 0.12 14.23 0.07 12.46 10.37 0.56
1500-1600 13.87 13.76 5.41 0.33 12.69 0.20 5.61 13.72 3.56
1600-1700 2.06 21.01 0.80 0.09 5.29 0.05 0.86 20.05 6.09
1700-1800 2.78 19.65 1.08 0.27 23.56 0.17 1.25 20.17 13.33
1800-1900 9.63 43.05 3.76 1.30 45.19 0.79 4.55 43.42 17.4
1900-2000 15.58 64.95 6.08 3.40 69.08 2.07 8.15 66.00 25.44
2000-2100 2.04 73.99 0.80 5.21 73.21 3.18 3.97 73.36 79.99
2100-2200 2.63 70.67 1.03 10.27 76.94 6.26 7.29 76.05 85.9
2200-2300 1.91 83.88 0.75 42.95 81.72 26.19 26.94 81.78 97.23
2300-2400 0.85 80.54 0.33 28.13 86.38 17.15 17.48 86.27 98.11
>2400 0.08 89.45 0.03 7.93 83.80 4.84 4.87 83.84 99.34
Total 100.00 27.61 39.02 100.00 80.83 60.98 100.00 60.06
Table 3: Coal Beneficiation Results Summary (Batch Tests)
59
8.4. CONTINUOUS ANALYSIS OF DISCARD COAL SEPARATION
A final series of tests was conducted to determine the continuous operating characteristics of the
separator. The separator was run for 45 minutes with a feed flow rate of 18kg/hr. Again the split density
was 1996kg/m3. A separation inefficiency of 0.046 was achieved. With the separation results being seen
in Table 4 and Figure 41. From the results of this test work it is clear that the separator can function
continuously without detriment to the separation efficiency and with a continuous upgrading of the coal
from 60% ash to 28% ash content. The clean coal product could either then be further upgraded, or is
suitable for industrial use.
60
Density Floats Overflow/Feed Sinks Overflow/Feed Calculated Partition
Fraction Overflow Ash (%) (%) Overflow Ash (%) (%) Feedstock Ash (%) Coefficient
3
(kg/m ) (%) (%) Feed (%)
<1400 8.38 8.33 6.10 0.61 13.20 0.47 6.57 8.68 7.12
1400-1500 14.37 11.98 11.20 0.61 0.24 1.26 12.46 10.79 10.11
1500-1600 19.16 14.27 5.04 3.07 10.21 0.57 5.61 13.86 10.23
1600-1700 33.53 21.01 0.77 1.23 5.81 0.09 0.86 19.45 10.24
1700-1800 8.38 21.17 1.10 1.84 22.96 0.15 1.25 21.38 11.98
1800-1900 2.99 43.93 3.88 7.98 47.24 0.67 4.55 44.42 14.78
1900-2000 7.78 66.88 6.13 7.36 72.37 2.03 8.15 68.25 24.86
2000-2100 2.40 70.31 0.84 13.50 73.06 3.14 3.97 72.48 78.97
2100-2200 1.20 81.65 0.94 27.61 75.51 6.35 7.29 76.31 87.11
2200-2300 0.60 72.79 1.07 20.86 81.22 25.87 26.94 80.88 96.03
2300-2400 0.60 80.08 0.52 7.36 82.63 16.96 17.48 82.56 97.01
>2400 0.60 89.09 0.03 7.98 84.82 4.84 4.87 84.84 99.34
Total 100.00 24.61 37.61 100.00 70.48 62.39 100.00 60.06
Table 4: Coal Beneficiation Results Summary (Continuous Tests)
61
8.5. ACCUMULATION WITHIN THE SEPARATOR BED
Accumulation within any process is important to quantify. In the case of dense medium separation
accumulation will affect not only the bulk density of the bed but may also interfere with the fluidisation.
The density tracers did not pose a threat to accumulation due to their size. Coal on the other hand is brittle
and regularly chips and breaks especially when in the environment of constant collisions that is usually
experienced in jig separation. The fluidised separator provides an environment that produces far less
attrition than traditional methods. After all the batch and test runs no coal fragments were found to have
passed through the separator and into the bed. This additionally meant the bed operated with little or no
coal dust formation, and important health and safety consideration.
62
9 PART 2 : CONCLUSIONS
As a result of the test work conducted on the separator from initial optimisation through batch test work
and final continuous operation the following conclusions were reached.
Magnetite powder provides an effective separation medium that can be fluidised adequately by
air to create a stable separation density between 1850 and 2100kg/m3. An optimum separation
density was found to exist at 1996kg/m3.
The novel separator was found to operate under optimum conditions at a superficial gas velocity
of 0.13m/s and a separator rotation speed of 4revs/min.
The separation inefficiency Ep achieved was 0.057 for the runs involving the density tracers. The
split density was achieved at 1996kg/m3.
The average separation inefficiency Ep achieved was 0.046 for the batch runs involving the
Waterberg coal, which is comparable with traditional wet separation techniques. The 2.5kg
batches of coal were fed into the separator and allowed to separate over a period of 9 minutes.
The coal entered at an average ash content of 60.1%. 39.8% of the coal reported to the floats with
a final average ash content of 28.5%. The remaining 60.8% of the coal reported to the sinks with
a final ash content of 80.9%. The split density was achieved at 1996kg/m3.
The average separation inefficiency Ep achieved was 0.046 for the continuous runs involving the
Waterberg coal, which is comparable with traditional wet separation techniques. The coal was
separated at a capacity of 18kg/hr. The coal was fed into the separator at an average ash content
of 60.1%. 39.7% of the coal reported to the floats with a final average ash content of 24.6%. The
63
remaining 60.3% of the coal reported to the sinks with a final ash content of 76.41%. The split
density was achieved at 1996kg/m3.
The novel separator itself provided an effective mechanism by which the floats and sinks of the
separation may be removed from the fluidised bed. This separator not only had a minimal impact
on the stability of the bed, but also was highly effective preventing the accumulation of
middlings and providing a mixing action within the bed to prevent the formation of dead zones.
The scale-up potential of the bed is enormous, the advantages of dry beneficiation are clear, and
the operation of this bed has proved that the disadvantages traditionally associated with dry
beneficiation can be recognised and overcome.
64
10 Future Work
10.1. SCALE UP
The success of the lab scale separator has prompted investigation into the construction of a scaled up pilot
plant. This pilot plant will correct some of the problems discovered with the operation of the laboratory
scale equipment whilst allowing for large scale continuous operation. The idea was to increase the
capacity of the equipment by a factor of 35 and at the same time iron out any bugs from the previous
design. The final proposed future design would measure 1.5m in width by almost 6m in length. The
proposed design of this separator is discussed here however the final construction and operational tests
will not be considered in the scope of this dissertation.
The bed was lengthened in order to provide a greater residence time for the particles and thus increase
efficiency by decreasing the possibility of entrainment.
The plenum differed from the previous design in that it was lengthened to account for the new shape and
provided with four separate air supplies. These air supplies would need to provide approximately 45 times
the volume of air needed to run the smaller separator. The addition of separate supplies would mean that
air distribution though the length of the bed could be altered to examine the affects of fluidization quality
and bulk density variation throughout the beds length.
The removal system of the previous separator had been adequate but had occasionally shown itself to
have several traits that could be improved upon. The angled lifters that fed the chutes at the open ends of
65
the separator basket did not provide as speedy a removal as was always needed there appeared to be
congestion of particles at the chute entrance which occasionally caused blockage and unnecessarily
increased residence time, thus reducing the capacity of the separator. The improved removal system
consisted of angled lifters that dropped the floats and sinks onto outwardly moving conveyers which
would ensure the rapid removal of the particles. These conveyers would be mounted above a central
support plate that would run down the length of the separator.
The scraper for dragging the floats from the sinks removal section to the floats removal section would be
replaced by a more efficient design that would be suspended from below the central support plate. This
would ensure rapid transfer of the floats to the floats removal section of the bed.
The previous drive mechanism did place a fair degree of strain on the separator basket, and would not be
an optimal choice for the scaled up bed. Rather the basket would be suspended on rollers situated just
above the wall of the bed.
The materials used to construct the bed would need to be considerably more resilient than those needed
for the previous design in order to cope with the substantially increased weight of the magnetite medium.
Steel was recommended for the beds internals and supports and test work needs to be conducted to ensure
the load bearing capacities of the support design.
66
Removal
Conveyers
Driving
Rollers
Central
Support Plate
Figure 42: Isometric View of the Proposed Design for the Scaled Up Separator
Floats Scraper
Honeycomb
Support
Structure
Primary
Plenum
Figure 44: Sectioned Isometric View of the Proposed Design for the Scaled Up Separator
67
Figure 44: Construction Isometric View of the Proposed Design for the Scaled Up Separator
68
69
Figure 45: Isometric, Top, Front and Side Schematic Views of the Proposed Design for the Scaled Up Separator
11. REFERENCES
1. Chiang S.H. and Cobb J. T., 1993, Coal Conversion Processes, Cleaning and Desulphurisation,
John Wiley & Sons
2. Clark K. 1997 , The Business of Fine Coal Tailings Recovery, The Australian Coal Review July
1997.
3. Cleaner Coal Technology Programme, 2001 , Technology Status Report 015 Coal Preparation,
Department of Trade and Industry
4. Coulson J.M. and Richardson J.F., 2002, Chemical Engineering-Volume 2: Particle Technology
and Separation Processes, 5th Ed, Butterworth-Heinemann
5. Creamer M., 2009-03-20, Clean Coal Vital to Sustain 90%-Dependent South Africa, Says Wits
Academic, Mining Weekly
7. Donnelly J., 1999, Potential Revival of Dry Cleaning of Coal, The Australian Coal Review, Issue
8, 26-30
9. van Houwelingen J. A. and de Jong T. P. R. 2003, Dry Cleaning of Coal: Review, Fundamentals
and Opportunities. Geologica Belgica (2004) 7/3-4: 335 -3423
10. Industrial Technologies Program 2006, Development of a Novel Dry Processing Technology, U.S.
Department of Energy
11. Kunii D. and Levenspiel O., 1991, Fluidisation Engineering, 2nd Ed, Butterworths
12. MacPerson S.A. et al, 2009, Density Based Separations in the Reflux Classifier with an air-sand
dense-medium and vibration, Minerals Engineering 23 (2010) 74-82
13. Pell M., 1990, Gas Fluidisation. In: Handbook of Powder Technology, Elsevier
70
14. Perry R.H. and Green D., 1998, Perrys Chemical Engineers Handbook,7th Ed, McGraw-Hill
15. Qing-ru C. and Hai-feng W. 2006 , Clean Coal Processing and Utilization of Coal, The Chinese
Journal of Process Engineering Vol 6 No. 3
16. Turton A., 2008, Three Strategic Water Quality Challenges that Decision-Makers Need to Know
About and How the CSIR Should Respond, CSIR Report No. CSIR/NRE/WR/EXP/2008/0160/A
17. Young HD. and Freedman RA., 1996, University Physics 9th Ed. Extended Version with Modern
Physics, Addison-Wesley
18. Zhenfu L. and Qingru C., 2001, Dry Beneficiation Technology of Coal with an Air Dense-
Medium Fluidized Bed, International Journal of Mineral Processing 63: Issue 3, 147-175
19. Zhenfu L. and Qingru C., 2001, Effect of fine coal accumulation on dense phase fluidized bed
performance, International Journal of Mineral Processing 63: Issue 4, 217-224
21. Council for Geoscience, Coal Fields of the Republic of South Africa, www.geoscience.org.za
(http://www.geoscience.org.za/index.php?option=com_content&task=section&id=26&Itemid=44
1), (accessed 21/02/2008)
22. Department of Environmental Affairs and Tourism, Mean Annual Rainfall of South Africa,
www.environment.gov.za (http://www.environment.gov.za/enviro-
info/sote/nsoer/general/about.htm), (accessed 21/02/2008)
71
(http://free.financialmail.co.za/cgi-bin/pp-email.pl)
(http://free.financialmail.co.za/08/1128/cover/coverstory.html), (accessed 28/11/ 2008)
29. Tangshan Shenzhou Machinery Co., Ltd, FGX compound dry cleaning machine,
www.tsshenzhou.com (http://www.tsshenzhou.com/eng_cpzs.cfm), (accessed 20/03/2009)
72
12 APPENDICES
RUN ID : 20080422
Separator Gas Velocity v (m/s) 0.17
Separation Inefficiency Ep 0.013
Separator Rotation Speed
0.00
(rpm)
Number of Tracers Used 45 (3 tracers in each weight fraction)
Number of Separation Runs 15.00
75 (75% pass point) 2038.08
25 (25% pass point) 2011.92
50 (split density) 2025.00
73
2.222 2000 1
97.778 2050 44
100.000 2100 45
100.000 2150 45
100.000 2200 45
100.000 2250 45
100.000 2300 45
100.000 2350 45
100.000 2400 45
100.000 2450 45
100.000 2500 45
100.000 2550 45
100.000 2600 45
100.000 2650 45
100.000 2700 45
100.000 2750 45
100.000 2800 45
Table 5: Raw Data Obtained from Batch Test Run with Density Tracers
74
2120 2110 2110 2130 2125
Bed Y- Axis
2120 2075 2075 2100 2125
Bed X - Axis
Table 6: Raw Data Showing the Bulk Density Variation within the Fluidised Bed in Plan View (kg/m3)
75
12.2. APPENDIX 2: PART 2 RAW DATA FOR TRACER SEPARATION
0.047 3
0.059 4
0.090 5
0.120 6
Table 7: Separation Inefficiencies Obtained At a Gas Velocity 0.09877m/s for Varying Rotation Speed
0.074 3
0.058 4
0.099 5
0.099 6
Table 8: Separation Inefficiencies Obtained At a Gas Velocity 0.12m/s for Varying Rotation Speed
76
Fluidising Gas Velocity (m/s) 0.13
0.050 3
0.063 4
0.120 5
0.127 6
Table 9: Separation Inefficiencies Obtained At a Gas Velocity 0.1264m/s for Varying Rotation Speed
0.098 3
0.134 4
0.153 5
0.163 6
Table 10: Separation Inefficiencies Obtained At a Gas Velocity 0.13037m/s for Varying Rotation Speed
77
RUN ID : 20080204
Separator Gas Velocity v (m/s) 0.10
Separation Inefficiency Ep 0.044
Separator Rotation Speed
3.00
(rpm)
Number of Tracers Used 45 (3 tracers in each weight fraction)
Number of Separation Runs 15.00
75 (75% pass point) 2059.72
25 (25% pass point) 1970.83
50 (split density) 2015.28
78
100.000 2300 45
100.000 2350 45
100.000 2400 45
100.000 2450 45
100.000 2500 45
100.000 2550 45
100.000 2600 45
100.000 2650 45
100.000 2700 45
100.000 2750 45
100.000 2800 45
Table 11: Raw Data Obtained From Continuous Separation of Density Tracers
RUN ID : 20080311
79
Separator Gas Velocity v (m/s) 0.10
Separation Inefficiency Ep 0.059
Separator Rotation Speed (rpm) 4.00
Number of Tracers Used 45 (3 tracers in each weight fraction)
Number of Separation Runs 15.00
75 (75% pass point) 2062.50
25 (25% pass point) 1944.64
50 (split density) 2003.57
80
100.000 2400 45
100.000 2450 45
100.000 2500 45
100.000 2550 45
100.000 2600 45
100.000 2650 45
100.000 2700 45
100.000 2750 45
100.000 2800 45
Table 12: Raw Data Obtained From Continuous Separation of Density Tracers
81
RUN ID : 20080314
Separator Gas Velocity v (m/s) 0.10
Separation Inefficiency Ep 0.080
Separator Rotation Speed
5.00
(rpm)
Number of Tracers Used 45 (3 tracers in each weight fraction)
Number of Separation Runs 15.00
75 (75% pass point) 2014.58
25 (25% pass point) 1854.17
50 (split density) 1934.38
82
100.000 2300 45
100.000 2350 45
100.000 2400 45
97.778 2450 44
100.000 2500 45
100.000 2550 45
100.000 2600 45
100.000 2650 45
100.000 2700 45
100.000 2750 45
100.000 2800 45
Table 13: Raw Data Obtained From Continuous Separation of Density Tracers
83
RUN ID : 20080318
Separator Gas Velocity v (m/s) 0.10
Separation Inefficiency Ep 0.114
Separator Rotation Speed
6.00
(rpm)
Number of Tracers Used 45 (3 tracers in each weight fraction)
Number of Separation Runs 15.00
75 (75% pass point) 2006.25
25 (25% pass point) 1778.13
50 (split density) 1892.19
84
100 2300 45
97.778 2350 44
100.000 2400 45
100.000 2450 45
100.000 2500 45
97.778 2550 44
100.000 2600 45
100.000 2650 45
100.000 2700 45
100.000 2750 45
100.000 2800 45
Table 14: Raw Data Obtained From Continuous Separation of Density Tracers
85
RUN ID : 20080325
Separator Gas Velocity v (m/s) 0.12
Separation Inefficiency Ep 0.077
Separator Rotation Speed
3.00
(rpm)
Number of Tracers Used 45 (3 tracers in each weight fraction)
Number of Separation Runs 15.00
75 (75% pass point) 1987.50
25 (25% pass point) 1737.50
50 (split density) 1862.50
86
100.000 2300 45
100.000 2350 45
100.000 2400 45
100.000 2450 45
100.000 2500 45
100.000 2550 45
100.000 2600 45
100.000 2650 45
100.000 2700 45
100.000 2750 45
100.000 2800 45
Table 15: Raw Data Obtained From Continuous Separation of Density Tracers
87
RUN ID : 20080401
Separator Gas Velocity v (m/s) 0.12
Separation Inefficiency Ep 0.053
Separator Rotation Speed (rpm) 4.00
Number of Tracers Used 45 (3 tracers in each weight fraction)
Number of Separation Runs 15.00
75 (75% pass point) 1943.75
25 (25% pass point) 1837.50
50 (split density) 1890.63
88
100.000 2350 45
100.000 2400 45
100.000 2450 45
100.000 2500 45
100.000 2550 45
100.000 2600 45
100.000 2650 45
100.000 2700 45
100.000 2750 45
100.000 2800 45
Table 16: Raw Data Obtained From Continuous Separation of Density Tracers
89
RUN ID : 20080408
Separator Gas Velocity v (m/s) 0.12
Separation Inefficiency Ep 0.125
Separator Rotation Speed (rpm) 5.00
Number of Tracers Used 45 (3 tracers in each weight fraction)
Number of Separation Runs 15.00
75 (75% pass point) 1943.75
25 (25% pass point) 1837.50
50 (split density) 1890.63
90
100.000 2350 45
100.000 2400 45
100.000 2450 45
100.000 2500 45
100.000 2550 45
95.556 2600 43
100.000 2650 45
100.000 2700 45
100.000 2750 45
100.000 2800 45
Table 17: Raw Data Obtained From Continuous Separation of Density Tracers
91
RUN ID : 20080415
Separator Gas Velocity v (m/s) 0.12
Separation Inefficiency Ep 0.115
Separator Rotation Speed (rpm) 6.00
Number of Tracers Used 45 (3 tracers in each weight fraction)
Number of Separation Runs 15.00
75 (75% pass point) 2007.50
25 (25% pass point) 1778.13
50 (split density) 1892.81
92
100.000 2350 45
100.000 2400 45
100.000 2450 45
97.778 2500 44
100.000 2550 45
97.778 2600 44
100.000 2650 45
100.000 2700 45
100.000 2750 45
100.000 2800 45
Table 18: Raw Data Obtained From Continuous Separation of Density Tracers
93
RUN ID : 20080422
Separator Gas Velocity v (m/s) 0.13
Separation Inefficiency Ep 0.059
Separator Rotation Speed (rpm) 3.00
Number of Tracers Used 45 (3 tracers in each weight fraction)
Number of Separation Runs 15.00
75 (75% pass point) 2028.29
25 (25% pass point) 1910.42
50 (split density) 1969.35
94
100.000 2350 45
100.000 2400 45
100.000 2450 45
97.778 2500 44
97.778 2550 44
100.000 2600 45
100.000 2650 45
100.000 2700 45
100.000 2750 45
100.000 2800 45
Table 19: Raw Data Obtained From Continuous Separation of Density Tracers
95
RUN ID : 20080429
Separator Gas Velocity v (m/s) 0.13
Separation Inefficiency Ep 0.044
Separator Rotation Speed (rpm) 4.00
Number of Tracers Used 45 (3 tracers in each weight fraction)
Number of Separation Runs 15.00
75 (75% pass point) 1945.83
25 (25% pass point) 1857.81
50 (split density) 1901.82
96
100.000 2350 45
100.000 2400 45
97.778 2450 44
97.778 2500 44
97.778 2550 44
100.000 2600 45
97.778 2650 44
100.000 2700 45
97.778 2750 44
100.000 2800 45
Table 20: Raw Data Obtained From Continuous Separation of Density Tracers
97
RUN ID : 20080506
Separator Gas Velocity v (m/s) 0.13
Separation Inefficiency Ep 0.100
Separator Rotation Speed (rpm) 5.00
Number of Tracers Used 45 (3 tracers in each weight fraction)
Number of Separation Runs 15.00
75 (75% pass point) 1979.17
25 (25% pass point) 1778.41
50 (split density) 1878.79
98
100.000 2350 45
97.778 2400 44
100.000 2450 45
100.000 2500 45
97.778 2550 44
100.000 2600 45
100.000 2650 45
100.000 2700 45
100.000 2750 45
100.000 2800 45
Table 21: Raw Data Obtained From Continuous Separation of Density Tracers
99
RUN ID : 20080513
Separator Gas Velocity v (m/s) 0.13
Separation Inefficiency Ep 0.120
Separator Rotation Speed (rpm) 6.00
Number of Tracers Used 45 (3 tracers in each weight fraction)
Number of Separation Runs 15.00
75 (75% pass point) 2013.75
25 (25% pass point) 1773.21
50 (split density) 1853.39
100
100.000 2350 45
100.000 2400 45
97.778 2450 44
100.000 2500 45
100.000 2550 45
100.000 2600 45
100.000 2650 45
100.000 2700 45
100.000 2750 45
100.000 2800 45
Table 22: Raw Data Obtained From Continuous Separation of Density Tracers
101
RUN ID : 20080520
Separator Gas Velocity v (m/s) 0.13
Separation Inefficiency Ep 0.133
Separator Rotation Speed (rpm) 3.00
Number of Tracers Used 45 (3 tracers in each weight fraction)
Number of Separation Runs 15.00
75 (75% pass point) 1985.94
25 (25% pass point) 1720.83
50 (split density) 1853.39
102
100.000 2350 45
93.333 2400 42
95.556 2450 43
91.111 2500 41
100.000 2550 45
100.000 2600 45
100.000 2650 45
100.000 2700 45
100.000 2750 45
100.000 2800 45
Table 23: Raw Data Obtained From Continuous Separation of Density Tracers
103
RUN ID : 20080527
Separator Gas Velocity v (m/s) 0.13
Separation Inefficiency Ep 0.113
Separator Rotation Speed (rpm) 4.00
Number of Tracers Used 45 (3 tracers in each weight fraction)
Number of Separation Runs 15.00
75 (75% pass point) 1980.60
25 (25% pass point) 1754.17
50 (split density) 1867.39
104
100.000 2350 45
100.000 2400 45
100.000 2450 45
100.000 2500 45
100.000 2550 45
100.000 2600 45
97.778 2650 44
97.7778 2700 44
100.000 2750 45
100.000 2800 45
Table 24: Raw Data Obtained From Continuous Separation of Density Tracers
105
RUN ID : 20080602
Separator Gas Velocity v (m/s) 0.13
Separation Inefficiency Ep 0.142
Separator Rotation Speed (rpm) 5.00
Number of Tracers Used 45 (3 tracers in each weight fraction)
Number of Separation Runs 15.00
75 (75% pass point) 1964.58
25 (25% pass point) 1680.36
50 (split density) 1822.47
106
100.000 2350 45
100.000 2400 45
100.000 2450 45
100.000 2500 45
100.000 2550 45
100.000 2600 45
100.000 2650 45
100.000 2700 45
100.000 2750 45
100.000 2800 45
Table 25: Raw Data Obtained From Continuous Separation of Density Tracers
107
RUN ID : 20080609
Separator Gas Velocity v (m/s) 0.13
Separation Inefficiency Ep 0.114
Separator Rotation Speed (rpm) 6.00
Number of Tracers Used 45 (3 tracers in each weight fraction)
Number of Separation Runs 15.00
75 (75% pass point) 2006.25
25 (25% pass point) 1778.13
50 (split density) 1892.19
108
100.000 2350 45
100.000 2400 45
100.000 2450 45
100.000 2500 45
100.000 2550 45
100.000 2600 45
100.000 2650 45
100.000 2700 45
100.000 2750 45
100.000 2800 45
Table 26: Raw Data Obtained From Continuous Separation of Density Tracers
109
12.3. APPENDIX 3: PART 3 RAW DATA FOR COAL SEPARATION
111
12.4. APPENDIX 4: DENSITY TRACER SPECTRUM
112
2750 15a 6
2800 15 6
2850 16a 6
Table 31: Density Tracer Spectrum
113